The multi-session experience in Customer Service

A contact-centre agent handles many things at once — a chat in progress, an open case, a phone call about to start, knowledge being searched, customer history being checked. The Customer Service Workspace (formerly Omnichannel for Customer Service) is built around this reality: a multi-session, multi-tab experience where the agent can hold parallel conversations and records without losing context.

Sessions. Each active conversation or focused work item is a session. The agent can have multiple sessions open concurrently, switching between them like browser tabs. A session typically anchors on a customer interaction — a chat, a phone call, a case, an email being composed — and holds the relevant tabs alongside.

The session pane. Sessions appear in a vertical list on the left of the screen. Active session is highlighted; idle sessions show a count of unread messages or pending actions. Closing a session ends it.

Tabs within a session. Each session can have multiple tabs — typically the conversation (chat/voice/email), the customer record, the case, related knowledge articles, and any inspection or form the agent is filling. The tabs are independent — the agent can switch tabs without losing place in the conversation.

The productivity pane. A right-side pane in each session holds productivity tools — knowledge search results, smart assist suggestions, conversation summary, related cases, suggested replies. The pane is contextual: opening the conversation tab shows conversation-related tools; opening the customer record shows customer-related tools.

Smart Assist and Copilot. Real-time AI-driven assistance:

• Conversation summary — when the agent opens a chat already in progress (e.g. transferred from another agent), Copilot summarises what's happened in a sentence.
• Suggested replies — based on the conversation, AI suggests answer phrasings the agent can accept, edit, or reject.
• Knowledge surfacing — articles relevant to the current conversation appear without explicit search.
• Case summarisation — at handoff or escalation, Copilot drafts a structured case summary the agent reviews and saves.

Capacity profiles. Multi-session works only if assignment respects agent capacity. Each agent has a capacity profile declaring how much work they can hold simultaneously — typically 1 voice call, 3 chats, 5 cases, no email. Unified Routing honours the profile: a chat won't route to an agent already at the chat cap.

Channel-specific behaviour.

• Voice calls — exclusive: an agent on a live voice call can browse other sessions but the voice conversation is the focus.
• Chats — concurrent: multiple chats run in parallel.
• Emails — long-running: emails can be composed across many short sessions without losing draft state.
• Cases — background: cases sit as sessions while the agent works on conversations, then come to focus when needed.

Personalisation. Agents can configure their multi-session experience — number of tabs to remember, default productivity-pane tools, notification preferences. Supervisors can set defaults per role.

Performance. The multi-session UI is data-heavy on the client. Slow networks degrade the experience; the configuration of how many sessions auto-load, how aggressively the productivity pane pre-fetches data, and which tools are enabled matters for agent satisfaction.

Supervisor view. Supervisors see real-time dashboards of agent sessions, conversation health, escalation flags. Live conversation monitoring (whisper, barge, listen on voice; observe on chat) lets supervisors intervene on stuck calls.

Configuration discipline. Tune the multi-session experience for each agent role. A first-line chat agent's workspace differs from an escalation specialist's. Don't ship one configuration to all agents.

Operational reality. Once agents adapt to multi-session, they don't go back. Average handling time drops, customer satisfaction improves, agent satisfaction often improves too. The configuration investment is real but one-time.